Multilingual Topic Models for Unaligned Text

TL;DR

This paper introduces MuTo, a probabilistic multilingual topic model that discovers shared topics and matches between documents in different languages without requiring aligned or parallel corpora.

Contribution

The paper presents MuTo, a novel multilingual topic model that uses stochastic EM to find language matches and shared topics in unaligned multilingual texts.

Findings

MuTo effectively finds shared topics in real-world multilingual corpora.

MuTo successfully pairs related documents across languages.

The model operates without the need for parallel corpora.

Abstract

We develop the multilingual topic model for unaligned text (MuTo), a probabilistic model of text that is designed to analyze corpora composed of documents in two languages. From these documents, MuTo uses stochastic EM to simultaneously discover both a matching between the languages and multilingual latent topics. We demonstrate that MuTo is able to find shared topics on real-world multilingual corpora, successfully pairing related documents across languages. MuTo provides a new framework for creating multilingual topic models without needing carefully curated parallel corpora and allows applications built using the topic model formalism to be applied to a much wider class of corpora.